1 Questions

how many image files did we get from the database?
how many image files did we expect from the database?
does the results file match the expected number of results?

2 Data Analysis

2.1 VIS Results

In this analysis I only look at plant area as a first pass to make sure the extracted objects are within expectations. The output from the visible/RGB camera is found in two csv files in output/vis. Definitions for each of the phenotypes in these csv files can be found here.

2.1.1 Analysis of Image Files

Imaging started on 2019-11-14 and ended on 2019-11-21. There were 0 days without an image for any sample.

The experiment was run for 8 days with 42 samples, and 1 image per day per sample, so we expected 336 images. We have 327 images. This is 97.3%. Hence we are missing 9 or 2.7% of images.

We expect 42 images per day:

2.1.2 Analysis of Results

Here we will read in the output file from our plantcv image analysis. Hopefully each image was analyzed and has a corresponding output.

output = read_csv(here::here('output','vis','vis.csv-single-value-traits.csv'),
                  na = 'NA') %>% 
  mutate(jobdate = as.Date(timestamp))

gmap = read_csv('data/genotype_map.csv', col_types = list(col_character(),col_integer(), col_character()))

gtypesummary = gmap %>% 
  count(gtype)

output <- full_join(output,gmap, by = c('plantbarcode' = 'sampleid', 'roi'))

We expected 42 data points per day.

This graph shows the number of entries on each date in the output file. This includes NA values.

We can check to see if we lost any output data compared to the input. For example, if a plant died then we’d see a step drop in the number of samples. The differences on each day between the graph above and the graph below indicates the number of lost plants.

2.1.2.1 Replication

Do the datapoints from the different replicates look stable on each day?

We need to test if all the ROI are present and distinct.

Table: Sample IDs with missing ROI (shown in black below)

Table: Sample IDs with nondistinct ROI (shown in red below)

0 data points with multiple nondistinct roi will be removed from the dataset and saved as level1 product. Please use the level1 output for further analysis.

2.2 PSII Results

In this analysis I only look at Fv/Fm and Y(II) for each induction period of the induction curve. The results are in output_psII_level0.csv and consist of Y(II), including Fv/Fm, and NPQ. Both the mean and standard deviations are reported.

2.2.1 Analysis of Image Files

Imaging started on 2019-11-14 and ended on 2019-11-22. There were 0 days without an image for any sample.

The experiment was run for 9 days with 42 samples, and 34 image per day per sample, so we expected 12852 images. We have 12852 images. This is 100.0%. Hence we are missing 0 or 0.0% of images.

We expect 1428 images per day:

2.2.2 Analysis of Results

Here we will read in the output file from our plantcv image analysis. Hopefully each image was analyzed and has a corresponding output.

n_param = 15 #there are n photosynthetic parameters -->

output2 = read_csv('output/psII/output_psII_level0.csv',
                  na = 'nan',
                  col_types = list(gtype = col_character(), roi = col_integer(), imageid = col_integer(), datetime = col_datetime(), jobdate = col_date(), parameter = col_factor())) %>% 
  mutate(
         gtype = parse_factor(toupper(gtype),levels=gtypeLevels),
         gtype = forcats::fct_relevel(gtype, 'WT',after=0))

imageid_map = read_csv(here::here('data/pimframes_map.csv'), col_types = list(col_integer(), col_character(), col_factor())) %>% 
  filter(!grepl('Abs',parameter)) %>% 
  droplevels()

output2 <- output2 %>% left_join(imageid_map)

We expect 1260 datapoints per day.

We can check to see if we lost any output data compared to the input. For example, if a plant died then we’d see the # of datapoints decrease over time.

Table: Number of result rows for each genotype/treatment

We expect

datapoints per day per genotype.

2.2.2.1 Replication